X-Ray Analysis of Foundry Dusts for Quartz and Iron in Relation to Silicosis and Siderosis Diffraction Analysis for Silica G. L. CLARK, W. F. LORANGER', and S. J. BODNAR2 Department o f Chemistry and Chemical hgineering, University o f lllinois, Urbana,
The quantitative x-ray diffraction analysis of industrial dusts, especially for a-quartz which is responsible for silicosis, and x-ray fluorescent-spectral analysis for iron are proved to be of immediate practical value i n the study of t h e dust hazards and ventilAtion efficiencies i n every part of a large steel foundry. For the first time all the specific foundry processes can be compared as to their potential hazard and can be correlated with actual diagnoses from lung radiographs of workers exposed u p to 20 years i n the various areas, i n terms of silicosis from a-quartz and of siderosis from iron. The original x-ray diffraction method of Clark and Reynolds is improved and adapted t o two different recording' Geiger diffractorneters with a linear proportionality between quartz concentration a n d corrected intensity ratios in each case. Two sets of dust samples collected a year apart were analyzed. The maximum a-quartz concentration was 81% for dusts in a molding area on down to 0% i n a grinding area i n which special new ventilation proved effective. Silicotic nodules were diagnosed i n many cases of extended exposure. The fluorescentspectral analysis for iron i n t h e dusts is proved to be far more reliable and convenient t h a n chemical analyses. I n other cases lung shadows were diagnosed as benign siderosis which was correlated with iron concentrations ranging from 0 to 30% i n t h e dusts. It is significant t h a t in this investigation x-rays were used i n three different ways-for diffraction analysis of a crystalline constituent, for spectral analysis of a noncrystalline constituent, and for radiographic diagnosis of pneumoconiosis. The typical results demonstrate t h a t entire industrial plants may be systematically and rapidly surveyed for dust hazards by parallel analyses and radiographic indications. The data were used as t h e basis of improving foundry practices and adding and improving ventilation, and for medical prescriptions for powdered-aluminum therapy or occupational change.
I
NDUSTRIAL dusts from mines, factories, foundries, quarries, etc., are known to be complex mixtures of siliceous and other compounds, both crystalline and amorphous. When such dusts contain silica, particularly a-quartz, they are considered a definite health hazard. The inhalation of dusts containing silica in appreciable amounts over a period of time may lead to silicosis, diagnostically recognized as nodules in lung radiographs. There is some controversy as to whether other forms of silica, such as glass or amorphous silica, cristobalite, and tridymite, may also be considered health hazards. The various silicates except asbestos which may be present appear to have benign effects on lung tissue. In order to study and control silicosis, methods of analysis yielding aclcurate quantitative and qualitative data are needed to distinguish the various forms of silica from other siliceous materials ( 5 ) . 1
Present address, United States Military ricademy. West Point, N. Y. address, Standard Oil Co of Louisiana, Baton Rouge, La.
* Present
'
111.
According to Goldman ( 3 ) ,routine wet chemical methods for the determination of quartz in industrial dusts give the quartz content of the sample only by a series of indirect calculations, for which it is necessary to identify petrographically or to assume the identity of the minerals other than silica present. This is an empirical method and the calculations usually lead to highly uncertain results, because when the particle size of the dust particles of many common silicate materials falls below 5 microns, identification is uncertain (3). This is an important aspect as the emphasis is now centered on the respirable dusts, those with particle sizes below 5 microns and more particularly below 3 microns. This dust of smaller particle size is most easily dissolved and lost in the chemical separations of silicates from silica and cannot be determined by the petrographic method ( 5 ) . Dusts with high iron content upon continued respiration will produce shadows on lung radiographs, diagnosed as siderosis, which is supposedly benign in respect to respiratory pathology. Rapid and reliable analysis of dusts in steel foundries for both crystalline quartz and iron, as metal or in compounds, and in crystalline or amorphous form is needed for correlation with medical observations. Quartz and iron analysis and medical diagnosis can be accomplished most successfully by three different x-ray techniques. THEORY
The identification and quantitative determination of crystalline constituents of a dust and the ultimate correlation of the results with the appearance or absence of silicosis of the lungs are useful and unique applications of x-ray diffraction analysis. A diffraction pattern may serve for quantitative as well as qualitative analysis. The relative intensity of the pattern of a substance in a mixture is proportional (except for certain absorption effects) to the amount present. Clark and Reynolds ( 1 ) in 1936 developed a method, adapting the internal standard method used in optical spectroscopy, for quantitative determination of quartz in mine dusts, which remained a standard procedure for a number of years. They recorded their diffraction patterns on a film and measured the density (photographic darkening on the film) of the pertinent lines. This line density was related to line intensity and working curves relating intensity to concentration were prepared. From this, they p-ere able to calculate the concentration of quartz in mine dusts. However, the advent of the Geiger-counter x-ray diffractometer eliminates the use of film and of photometric evaluation of densities, with consequent great saving in time, and provides automatically recorded intensity measurements of high accuracy, which are necessary in quantitative work. The application of the recording x-ray diffractometer to analysis of crystalline dusts was investigated and greatly r e fined by Klug and coworkers ( 5 ) . They also employed the internal standard method of Clark and Reynolds ( 1 )and prepared working curves in which the ratio of the intensity of the 3.35 A. quartz line to the 3.16 A. calcium fluoride line was plotted against concentration of quartz. Schmelzer (6) objects to the internal standard technique, since the internal standard and quartz lines are in separate regions and therefore subject to 1413
1414
ANALYTICAL CHEMISTRY
variations in intensity because of power fluctuations. However, xork a t the Rlellon Institute ( 5 ) and the investigations reported in this paper indicate the value in the use of an internal standard. The alternative technique employing measurements of absorption coefficients and appropriate formulas was perfected after the completion of the present investigation. EXPERIME*TAL
Two completely different sets of samples viere collected, one about a year after the other, by two different filtering techniques. These were analyzed on two different Geiger diffractometers by two research men working entirely independently. Of 20 dust samples, 16 were collected from specially designated zones in a large typical steel foundry by drawing the air through filters a t constant rate by means of a special vacuum apparatus for a stated period of time. From the total weight of sample it was possible to calculate the grams of dust collected per hour in the same volume of air, and from the dust analysis, the grams of a-quartz per hour. Four of the twenty samples were collected by hand from rafters in various zones of the foundry. A schematic diagram of the floor plan indicating the sites of sampling is given in Figure 1. Preparation of Standard S e r i e s . The method of Clark and Reynolds ( 1 ) was followed in the preparation of a standard series for the working curves, from a-quartz, which analyzed better than 99.9% pure, equally pure amorphous silica used as a diluent, and calcium fluoride used as the internal standard. Each of the 16 samples for the standard series with percentages of quartz varying from 2 to 100% was made up to 2.50 grams. Each sample contained 0.50 gram of calcium fluoride, or 20% of the total weight of each sample. and amorphous silica, which gave no diffraction maxima between 5 and 8 5 O , was used as the diluent in requisite amounts. After the samples were weighed into screwtop vials, they were intimately homogenized on a mixing wheel for 21 hours. In preparing the sample for the holder to place on t'he x-ray Geiger-count'er unit. the entire sample (2.50 grams) was placed on a 6inch-square piece of black glazed weighing paper on a flat surface. The quadrant system of sampling was then used t o reduce the size of the sample to an amount. sufficient to fill a Lucite holder 1 X 2 cm. X 2 mm. in depth. In this lvork it \vas again demonstrated that mixing, sampling. arid packing of the sample in a suitable holder usually is the predominant source of error in a quantitative analysis-by the use of x-rays. Preferred orientation of powder particles can cause abiiornially l o x intensities of diffraction maxima and eYen cause the disappearance of some lines altogether ( 5 , 6). Care must be taken in packing a powdered crystalline sample of a particle size of about 5 microns to ensure that the distribution of the particles is as random as possible. For the first set of samples. the'sample was pressed into the rect,angular Lucite cell with the spatula blade until a smoot,h surface was obtained. Then the edge of the blade of the spatula was drawn in a sxveeping motion Over t'he surface of the sample
1415
V O L U M E 2 6 , NO. 9, S E P T E M B E R 1 9 5 4 to give it a rough appearance. Schmelzer (6) independently employed such a met,hod. It was finally possible to reproduce the sample preparation so that in repeated measurements intensities of diffraction maxima, both for quartz and for calcium fluoride, did not vary appreciably from run to run. For the first set of samples the Norelco 90" diffractometer was used and for the second a new General Electric XRD-3 unit. In the first instrument the sample is horizontal; in the second i t is vertical. More difficulty was encountered with the latter arrangement in reproducing intensity results because of variations in packing the powders. Best results finally were obtained with a larger sample in an aluminum container with a cavity of 0.5 X 1 5 x 1/16 inches, bathed by a beam collimated by the largest slits The powder was firmly pressed into the cavity and the elceis scraped off with the edge of a spatula held a t a 45" angle to the surface. X-Ray Analysis of Standard Series. The first set of dust samples was analyzed on the Sorelco diffractometer with a copper target operated a t 35 peak kilovolts and 6 ma. The counter was set at 60" on the analyzer and moved a t a constant rate of 2" per minute until the counter reached 12'. Therefore, total exposure tinic per pattern was 24 minutes. An excellent base line was provided by running a pattern of amorphous silica on the same pattern on which the sample is recorded. Table I lists the corrected intensity values (peak height minus the background in arbitrary units on the recorder chart which may be calibrated in counts per second) for the 3.35 A. line of quartz and 3.16 A. line of calcium fluoride, as well as the intensity ratio of quartz to calcium fluoride for each of the samples in the standard series. -
-~
~
-
- _ -
Cndoubtedly all the set I1 data deviate less from a straight line. The discrepancy must be either in the packing of the horizontal and vertical samples, respectively, in sets I and 11, or in the difference in measuring and correcting intensities of linear peak heights on recorded charts in set I and in counts per second (from times for fixed total counts) in set 11, respectively. In either case greatest difficulty was always encountered in reproducibility for quartz samples, probably owing to varying tendency toward preferred orientation of the powder grams.
__
~
F E Q C E k T QUAQTZ
Table I. Standard Series Measurements for Working Curves on Corrected Intensity Ratios of 3.35 A. Line of Quartz to 3.16 A. Line of Calcium Fluoride (ZQ/ZC) Quartz,
9%
0 2.5 5.0 7.5 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100 0
A r . rq ' I C Set I , Set 11, Norelco-peak G.E.-Geiger heights on charts counts 0.20 ... 0.51 0.62 0.95 i.31 1.17 3.25 1.77 4.34 3.13 6.05 4.74 7.41 5.13 6 . .53 8.53 8 00 9.88 11.57 8.89 13 24 9.16 14 84 11.Il0
Av. Dev., Set I1
Av Dev.. c/o, Set I1
0.90 0.22 0.20 0.31 0.40 0.45 0.21 0.80 1.12 1.15
14 9 (8
8
4 ti 5.1 5.4 .5 3 2.1 6 9 8 5 7.7
For the second set of dust samples analyzed on the General Electric unit, actual Geiger counts were made for the 3.33 A. peak of quartz with the scaler set a t the 16384 total count range, for the 3.16 A. peak of calcium fluoride with the scaler a t 2048, and for the background intensity a t 2048 or 1024. The times recluired for these respective rcadings then enabled calculation of counts per second. -1ctually three different procedures were critically compared. These involved different slit widths and scaler ranges for the three measurements, each on several repacked samples arid oil diflerent surface areas. The conditions which gave the best agreement, with a linear working curve were a detector slit width of 0.2", a beam defining slit of 3.0",and x-ray tube operating conditions of 34 kv. and 2 ma. The intensity ratios represent'ing thc average of five separate determinations with repacked samples, together with the average deviation and the percentage averagp deviation, are also tabulated in Table I.
Figure 2. Working Calibration Curves for Quantitative Determination of Crj-stallinea-Quartz in Dust Samples I Q ~ I ~Ratio , of intensities of 3.33 -1.line of a-quartz t o 3.16 A . line of CaF2 used a s internal standard 1. S o r e l r o diffractometer, horizontal specimen, peak heights on c h a r t 2 . CE diffrartornrter, \-ertical sample, Geiger counts a t peak positions
The r e 4 t s for standard samples in the range from 10 to 90% quwtz, with less than 2% error between actual and calculated values for each method, are as good if not better than can be expected in quantitative work using x-ray procedures. I n the low range of 0 to 10% quartz measured in set I analyses the procedure is somewhat less reliable, since ordinarily, the concentrations under 5% are difficult to determine either qualitatively or quantitatively by x-ray diffraction, even with film techniques. Preparation of Dust Samples. Since in preparing the standard aeries, 2 0 5 by weight of the sample was internal standard (0.50 gram of calcium fluoride in a total of 2.50 grams of sample), the same percentage by weight of calcium fluoride was added to each of the iinknown dust mixtures. S o diluent was added to any of the duyt samples, as there was sufficient quantity for a deter-
Table 11. ..lnalj-sisof Set I of ..iir-Borne Dust Samples from Foundry Sample SO.
%-1 Z.2 2-3 %-4
Preparation of Working Curves. \\-hen the intensity ratio oi the silica line to the fluorite line, as calculated from the specti,ometer charts for each sample of the standard series, is plotted against the knox7.n concentration of each member of the series, a straight line can be drawn extending through the origin, from 0 to 100yo quartz concentration for each of the two sets of iiidependentlg determined data. The two lines for set I (Norelco) and set I1 (for corresponding standard samples of the same composition) do not coincide. The working curve for set I1 lies above I and has a steeper slope so that the 100% quartz sample has an intensity ratio value nearly 4 units higher.
%-5 %.fi
z-7 %-Y
%-'3
%-10 %-I 1 L-1%
%-13 %-I4 Z-15
%-le
TTt.. G 3.6334 1 ,0850 0 7690 (i . 0000 1 .0930 0.60000 0 80000 3.8000 1.4700 2.8200 0.8550 0.7300 1.1150 3.1790 0.7700 1.5800
R- 1
n -2 H-8 H-4
..
ry I C 2.12 2 49 4.16 2 72 : i0 1 3 I0 0.8s 5.85 9.00 1.80 2.00 0 07 3 95 0 0 0 2.80 3.88 4 35 0
Quartz,
yo
19 R 22 5 37.7 24.8 27.2 28.G 7.1 .54 8 81.0 lii 5 1s 0 $ 5 36 0
0 0 0 2.5 5 33.3 41.5
0
Hours Collected 16 8 16 4 8 16 I(i 16 1ii
16 1t i
16 1t i
16 24 24
G. Quartz/ Hour 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0438 0205 0181 3720 0372 0105 0027 1301 0744 0290 0096 0025 0261
1416
ANALYTICAL CHEMISTRY
mination on the x-ray unit and the sample was recovered after each run. The amounts of calcium fluoride added to the dust samples varied from 0.12 (for the smallest sample of dust) to 1.71 grams (for the largest sample of dust). After adding the internal standard, the dust samples were tightly stoppered and revolved on a mixing wheel for 24 hours. It was unnecessary to resort to ball-mill methods for mixing or grinding.
Table 111. Analysis of Set I1 of Air-Borne Dust Samples from Foundry Sample No. z-1
Quartz,
7%
24.3 19.6 19.6 33.8 6.3 35.7 79.4
2-2 2-3 2-4 2-5 Z-6
E-%
e-9'
z-10 z-11 z-125 Z-135 2-14 Z-15a Z-16
a
Cu. Ft. Air Mg. Quartz Mg. Quartz/ Sampled/Min. Collected/Min. Cu. Ft. Sampled 67.1 20.8 0.309 80.1 5.53 0 .0690 82.1 2.01 0.0245 69.9 34.3 0.491 68.4 1.91 0.079 59.1 2.42 0.0409 66.4 43.8 0 . 660
....
....
35:s 51.7
7213 65.2
6.36 5.21
n.0880 0.0799
ca. i : 3
62:7
0,694
0.0111
i:7
6212
0,399
0.0641
..
..
.... .... ....
.... .... ....
Volume of sample not large enough for technique of analysis.
X-Ray Analysis of Dust Samples. The sample holders were packed in the same manner as were the standard series samples, after the dust samples had been reduced to suitable size, if necessary, by the quadrant method. The diffractometers were operated in the same manner as before. The first set of samples (21-216, Rl-R4) collected in 1952 from the areas indicated in
Figure 1 were analyzed with the Norelco diffractometer and the working curve I in Figure 2 was used to read off the percentages of quartz from the corrected intensity ratios, I Q / I ~ . These data are assembled in Table I1 together with the grams of quartz collected per hour. The second set of samples was collected in 1953 by a much improved method (Visi-Float), employing standard concentric ring paper filters, and was analyzed with the General Electric diffractometer. The quartz content read off from the working curve I1 in Figure 2. These data are assembled in Table I11 together with the cubic feet of air sampled per minute and the milligrams of quartz per minute and per cubic foot. Again it is emphasized that two sets of dust samples collected by two different methods a year apart, but in approximately the same foundry areas, were analyzed with two different diffractometers by two different observers using two different methods of measuring and correcting intensities of diffraction lines. Except for the direct experimental comparison of standard samples, indicating systematically higher values by the second technique, the data on dust comparison are consistent within each group but can give only approximately the change in the dust picture in the foundry during the course of a year. Samples R-1, R-2, R-3, and R-4 were hand-collected from places where dust had settled in rather large amounts. These samples contained up to 100 grams of dust. The dusts as received were placed in a Patterson-Kelley twin-shelled blender and mixed for 1 hour. The shell of the blender is of the V-type design which splits the sample and then recombines it every time the blender rotates. After homogenization, a portion of the dusts was ground until it all passed a 250-mesh screen. The dust was then weighed to the nearest milligram on an analytical balance and the proper amount of calcium fluoride was added as before. The dust and calcium fluoride were placed in I-ounce screw-capped bottles, placed on a mechanical mixer, and mixed for 4 hours.
(X-RAY ANALYSIS OF FOUNDRY DUSTS)
Fluorescent Spectral Analysis for Iron G. L. CLARK and H. C. TERFORD' Department o f Chemistry end Chemical Engineering, university o f Illinois, Urbana,
T
HERE has never been certain evidence of metallic iron or any of its crystalline compounds in the diffraction patterns of dusts even where 30% or more of the element is present. Because of the importance of iron determinations in foundry dusts in the diagnosis of siderosis, as contrasted with silicosis, and because of the extremely laborious and unsatisfactory analysis by most chemical methods (requiring fusions to get silicates into solution), recourse must be taken to fluorescent-spectral analysis. This is done with the same General Electric Geiger unit aa employed for quantitative diffraction analysis for a-quartz. 1
Present address, Shell Chemical Co., Houston, Tex.
111. APPARATUS
The General Electric SPG fluorescent spectrometer was used with the XRD-3 x-ray diffraction unit. The Machlett AEG-50 tungsten tube was operated a t 50 kv. and 50 ma. The fluorescent radiation was analyzed by diffraction through a mica crystal 0.0013 inch thick and defined by a 0.3" detector slit before the Geiger tube. The sample holder was of '/(-inch aluminum stock in which a cavity of I-inch diameter and 1/,-inch depth had been machined. When the holder was positioned in the sample drawer the sample completely covered the hole of the aluminum mask which defines the irradiated area. CALIBRATION CURVE
Table IV.
Composition of Standard Samples
SiOn, CaFn', Fe, G. G. % 1 3.792 0.948 1 3.696 0,924 2 3 0.900 5 3 3.600 4 0.880 3.552 6 0.852 5 3.408 9 6 0,600 0.840 1.200 3.360 10 7 0,900 1,200 3.120 0.780 15 0 The dust sample8 to be analyzed contained .20% CaF? which ,had been sdded as internal standard for the quartz determination by x-ray diffraction. Later work has proved that nickel as powdered metal or pure salts 8erves sa a satisfactory internal standard for both the diffraction and spectral snalraes. Sample
No.
Fe, G. 0.060 0.180 0.300 0.360 0.540
Ni, G. 1,200 i.2m 1.200 1.200 1,200
The calibration curve was established by plotting the ratio of the intensity of the iron K a to that of the nickel K a as a function of composition as determined from standard homogeneous samples of known composition. Table IV shows the composition of the standard samples, the constituents of which were passed through a 250-mesh sieve, weighed, and mixed for 2 hours in a Patterson-Kelley twin-shelled blender. The peak intensities of the iron K a and nickel Ka were scaled and the background for each was determined by scaling 2' on each side of the peak and taking the average value as background. The ratio of net intensities above background, I(FeKa)/
